Current File : //usr/share/doc/libgsm1/README.Debian
On December 20, I downloaded the following two emails about the patent
status of the GSM 6.10 algorithm.
Felix Lechner <felix.lechner@lease-up.com>
* * *
https://www.mail-archive.com/ietf@ietf.org/msg03978.html
From: James P. Salsman
Sent: Tuesday, November 14, 2000 3:03 pm
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: [VPIM] GSM 6.10 is public domain; audio/wav needs registered
Contrary to what people at VPIM meetings, lists, and on web pages
have suggested, nobody owns any IPR on the GSM 06.10 vocodec format,
or on any routines for encoding or decoding it. It was developed
from published code by people who took care to publish it before it
could be monopolized.
Philips owns the rights to a related but different form of LPC,
from U.S. patent 5,943,646, which was applied for more than four
years after the publication of GSM 06.10 by ETSI. That patent is
most likely what is confusing people about the status.
Also, http://www.ema.org/vpimdir/specs/draft-ema-vpim-wav-00.txt
-- the pending-in-limbo audio/wav IANA registration -- has an error:
it reads "audio/vnd.wav"; that should be "audio/vnd.wave", which,
by the way, hasn't been registered with IANA either. I agree with
Keith Moore that they should be registered as identical, and I hope
that they be registered in the same document to make that clearer.
The definitive reference for these formats seems to be kept in
Microsoft's Support Knowledge Base Article ID Q120253:
http://support.microsoft.com/support/kb/articles/Q120/2/53.asp
Cheers,
James
* * *
https://www.mail-archive.com/ietf@ietf.org/msg03978.html
Re: [VPIM] GSM 6.10 is public domain; audio/wav needs registered
James P. Salsman Wed, 15 Nov 2000 15:56:50 -0800
Jutta,
Thanks for the information:
> The patent I've seen investigated in connection with GSM 06.10
> and Philips is the older 4,932,061 (1990)....
Interesting. The priority date of that one is 22 March 1985.
The practice of quantizing residual exitation in LPC vocoders was
not novel in 1985. For an example of how people were performing
VQ classification on the exitation residual much earlier, see:
"Epoch extraction from linear prediction residual for
identification of closed glottis interval", by T. Ananthapadmanabha
and B. Yegnanarayana, in IEEE Transactions on Acoustics, Speech,
and Signal Processing, vol. 27, no.4, pp. 309-19 (1979).
Certainly the methods of doing such quantization described in the
claims of patent 4,932,061 are novel, because they all explicitly
refer to "perceptually weighting" the exitation residual. However,
GSM 6.10 uses only four quantization vectors and a linear scaling
factor, without any weighting based on non-linear perceptual
modeling, so that particular set of claims do not apply to GSM 6.10.
It also might be helpful to look at the references in the these
GSM 6.10 descriptions published prior to ETSI's:
"Evolution of Six Medium Bit rate Coders For The Pan-European
Digital Mobile Radio System", by E. Natvig, in Journal On Selected
Areas In Communications, vol. 6, no. 2, pp 324-34 (1988).
"Speech Codec for the European Mobile Radio System", by K. Hellwig,
P. Vary, D. Massaloux, and J.P. Petit, in Proceedings of the
ICASSP-88, pp. 227 (April, 1988).
Cheers,
James
Mini Shell